{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:LTIA47SD7O5VD7S3PMFMZ67QSS","short_pith_number":"pith:LTIA47SD","schema_version":"1.0","canonical_sha256":"5cd00e7e43fbbb51fe5b7b0accfbf094856d4ef0ae1f60c0974947ebe06ec660","source":{"kind":"arxiv","id":"2208.02813","version":1},"attestation_state":"computed","paper":{"title":"Towards Understanding Mixture of Experts in Deep Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Quanquan Gu, Yihe Deng, Yuanzhi Li, Yue Wu, Zixiang Chen","submitted_at":"2022-08-04T17:59:10Z","abstract_excerpt":"The Mixture-of-Experts (MoE) layer, a sparsely-activated model controlled by a router, has achieved great success in deep learning. However, the understanding of such architecture remains elusive. In this paper, we formally study how the MoE layer improves the performance of neural network learning and why the mixture model will not collapse into a single model. Our empirical results suggest that the cluster structure of the underlying problem and the non-linearity of the expert are pivotal to the success of MoE. To further understand this, we consider a challenging classification problem with"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2208.02813","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-08-04T17:59:10Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"a6fc029334adb845a75a4113d6b511750ff48565dc77798c0286e23b03861bff","abstract_canon_sha256":"7a5bca50df694056e93ef155fa3c02d6fc88106257ca838e6f8ed3469f610f51"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:46:09.297411Z","signature_b64":"JBriCSZJJoaqWTHTTGtQ1n2m9u5Ho+a10p8IeZe7gGsodn8XLOS1EppyandWkkJN9BFJ0a1NzvKXpM3EGZUdBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5cd00e7e43fbbb51fe5b7b0accfbf094856d4ef0ae1f60c0974947ebe06ec660","last_reissued_at":"2026-07-05T04:46:09.296942Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:46:09.296942Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Understanding Mixture of Experts in Deep Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Quanquan Gu, Yihe Deng, Yuanzhi Li, Yue Wu, Zixiang Chen","submitted_at":"2022-08-04T17:59:10Z","abstract_excerpt":"The Mixture-of-Experts (MoE) layer, a sparsely-activated model controlled by a router, has achieved great success in deep learning. However, the understanding of such architecture remains elusive. In this paper, we formally study how the MoE layer improves the performance of neural network learning and why the mixture model will not collapse into a single model. Our empirical results suggest that the cluster structure of the underlying problem and the non-linearity of the expert are pivotal to the success of MoE. To further understand this, we consider a challenging classification problem with"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2208.02813","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2208.02813/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2208.02813","created_at":"2026-07-05T04:46:09.297000+00:00"},{"alias_kind":"arxiv_version","alias_value":"2208.02813v1","created_at":"2026-07-05T04:46:09.297000+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2208.02813","created_at":"2026-07-05T04:46:09.297000+00:00"},{"alias_kind":"pith_short_12","alias_value":"LTIA47SD7O5V","created_at":"2026-07-05T04:46:09.297000+00:00"},{"alias_kind":"pith_short_16","alias_value":"LTIA47SD7O5VD7S3","created_at":"2026-07-05T04:46:09.297000+00:00"},{"alias_kind":"pith_short_8","alias_value":"LTIA47SD","created_at":"2026-07-05T04:46:09.297000+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.31010","citing_title":"MoG: Mixture of Experts for Graph-based Retrieval-Augmented Generation","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2508.12247","citing_title":"STM3: Mixture of Multiscale Mamba for Long-Term Spatio-Temporal Time-Series Prediction","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2508.12247","citing_title":"STM3: Mixture of Multiscale Mamba for Long-Term Spatio-Temporal Time-Series Prediction","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16349","citing_title":"Geometric Asymmetry in MoE Specialization: Functional Decorrelation and Representational Overlap","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17889","citing_title":"CoX-MoE: Coalesced Expert Execution for High-Throughput MoE Inference with AMX-Enabled CPU-GPU Co-Execution","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14200","citing_title":"How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19042","citing_title":"STK-Adapter: Incorporating Evolving Graph and Event Chain for Temporal Knowledge Graph Extrapolation","ref_index":68,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS","json":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS.json","graph_json":"https://pith.science/api/pith-number/LTIA47SD7O5VD7S3PMFMZ67QSS/graph.json","events_json":"https://pith.science/api/pith-number/LTIA47SD7O5VD7S3PMFMZ67QSS/events.json","paper":"https://pith.science/paper/LTIA47SD"},"agent_actions":{"view_html":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS","download_json":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS.json","view_paper":"https://pith.science/paper/LTIA47SD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2208.02813&json=true","fetch_graph":"https://pith.science/api/pith-number/LTIA47SD7O5VD7S3PMFMZ67QSS/graph.json","fetch_events":"https://pith.science/api/pith-number/LTIA47SD7O5VD7S3PMFMZ67QSS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS/action/storage_attestation","attest_author":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS/action/author_attestation","sign_citation":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS/action/citation_signature","submit_replication":"https://pith.science/pith/LTIA47SD7O5VD7S3PMFMZ67QSS/action/replication_record"}},"created_at":"2026-07-05T04:46:09.297000+00:00","updated_at":"2026-07-05T04:46:09.297000+00:00"}